首页> 外文期刊>Quality engineering >A Method for Identifying Which Tolerance Causes Malfunction in Assembled Products
【24h】

A Method for Identifying Which Tolerance Causes Malfunction in Assembled Products

机译:确定组装产品中哪些公差导致故障的方法

获取原文
获取原文并翻译 | 示例
       

摘要

Taguchi's example discussed in this column is an excellent illustration of a general and useful approach to quality improvement and defect reduction. Indeed, the method is not restricted to two fits but can easily be extended to any number and could be extended to the situation where the fits are measured, not just categorized. It is common to sample lots of material and check for defects. We do that, for example, when we make control charts for attributes (i.e., p, pn, c, and u charts). But if we also measure or characterize physical properties of the items we sample, and record covariates (x's in regression language) as Taguchi did, then we may be able to gain valuable insight into the possible causes of defects beyond just looking for patterns in time. Conjectures emanating from such studies, because that is really all it will be, can then be used in selecting factors to control in a formal factorial experiment to verify possible cause-and-effect relationships. In this column, we have deliberately demonstrated simple approaches for the statistical analysis based on Pearson's chi-square test. Such methods will most likely bring out the main features of the data. However, many new and powerful methods for the analysis of binary data have been developed over the years and increasingly are supported by major software packages―making them readily available and practical alternatives to the classic Pearson's chi-square test. One such method is logistic regression. For practical introductions to this subject, see e.g., Ref. [1] or Ref. [3].
机译:Taguchi在本专栏中讨论的示例很好地说明了提高质量和减少缺陷的通用且有用的方法。实际上,该方法不限于两个拟合,而是可以轻松地扩展为任意数量,并且可以扩展为测量拟合的情况,而不仅是分类的。通常对大量材料进行采样并检查缺陷。例如,当我们为属性制作控制图(即p,pn,c和u图)时,便会这样做。但是,如果我们也测量或表征所采样的物品的物理特性,并像田口一样记录协变量(回归语言中的x),那么我们不仅可以及时寻找模式,还可以深入了解缺陷的可能原因。 。这样的研究产生的猜想,因为那将是全部,因此可以用于选择要进行正式析因实验的控制因素,以验证可能的因果关系。在本专栏中,我们故意展示了基于Pearson卡方检验的简单统计分析方法。这种方法很可能会带出数据的主要特征。但是,这些年来,已经开发了许多用于分析二进制数据的强大的新方法,并且越来越受到主要软件包的支持-使它们易于使用,并且可以替代经典的Pearson卡方检验。一种这样的方法是逻辑回归。有关此主题的实用介绍,请参见例如参考资料[1]或编号。 [3]。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号